Relapse poses significant challenges to the treatment of a broad array of behavioral disorders, especially for addictions like cigarette smoking. Relapse can be considered a reversal in preference. Unfortunately, the behavioral phenomena that contribute to or underlie relapse are generally not well understood. One such process is delay discounting (DD), which refersto the reduced value or worth of a delayed reinforcer compared to the value of an immediate reinforcer. Importantly, DD predicts the preference reversals (PR) inherent in relapse by providing a mechanistic account for a change in preference. Preferences reversefrom the larger, more temporally distant reinforcer to a smaller, more immediate reinforcer as the immediate one becomes closer temporally. DD also predicts circumstances where the likelihood of PR are low; that is, Individuals with very low discount rates will tend to prefer larger, delayed reinforcers and those with very high discount rates will prefer smaller, sooner reinforcers.The understanding of PR are important because these reversals 1) provide a mechanism to understand relapse (a reversal from preferring abstinence to preferring a returnto the addictive activity) and 2) may explain the three distinct subgroups that collectively produce the typical relapse curve: that is, (a) the small number of individuals who do not respond to treatment or relapse immediately post-treatment, (b) the small portion individuals who do not relapse during the post-treatment measurement period and (c) the vast majority of individuals who respond to treatment, but later relapse at one of the post-treatment measurement periods. We propose to determine if discounting can predict relapse when viewed as a continuum and whether it predicts relapse when viewed categorically. Additionally, we will compare whether the continuum or categorical approach accounts for more of the variance associated with the data. We will also compare theoretically derived logistic models composed of DD, dependence, impulsivity, and negative affect measures as predictors of relapse and/or success following treatment. To complete these aims will collect pre-treatment measures of discounting and other measures and examine their relationship to relapse among tobacco smokers. This population is ideal because tobacco smoking is a major public health problem, tobacco smokers demonstrate considerable relapse and exhibit extreme discounting. Identifying these phenotypes and achieving these aims will have important theoretical and practical implications, including 1) clarifying our understanding of neurobehavioral mechanisms and processes that underlie relapse and its inverse-successfulabstinence-and 2) suggest novel ways to tailor treatments for enhanced outcomes among phenotypic subgroups.